You built the prototype with AI.Now make it a real product.
I help non-technical founders and operators turn apps built with Lovable, Bolt, or Cursor into secure, reliable software, or build the MVP right the first time.
The Vibecoder Ceiling
AI builds prototypes in minutes. Systems run businesses for years.
Every fix breaks something else
The AI keeps going in circles. You fix one bug, two more appear, and you're no longer sure which version of your app is safe to launch.
Real users expose the hard parts
Login loops, customers seeing each other's data, payments or emails that silently fail. The builder hid these problems until real people showed up.
Surprise bills and outages
One busy day and your API bill spikes or the app goes down. Nothing is watching costs, usage limits, or errors, so you find out from your users.
How We Work Together
Pick the engagement that fits your stage.
Whether your AI-built app is stuck, or you haven't started yet, there's a clear next step.
Prototype-to-Production Audit
Best for: Founders who built on Lovable, Bolt, Cursor, or Replit and hit a wall: a bug loop, a broken login, or a database that won't cooperate.
What you get
- A full review of your codebase and where it's fragile
- A security check, with exposed keys and credentials locked down
- A cleanup plan for your Supabase database
- A step-by-step roadmap to launch cleanly
Rapid MVP Sprint
Best for: Operators with clear wireframes, AI mockups, or business workflows who need a real MVP built right the first time.
What you get
- A production-ready Next.js + Supabase app
- Third-party integrations with cost controls built in
- Automated tests so new features don't break old ones
- Hands-off deployment on Vercel
Fractional Technical Operator
Best for: Seed-stage startups and solo operators who need an embedded product and technical ops lead without a full-time hire.
What you get
- Someone steering your AI development cycles
- Specs, backlog, and guardrails that keep AI tools on track
- Internal tools and automations for your operations
Not sure which fits? Email me what you're building →
Proof of Execution
Case Studies
Products I built and run myself, using the same AI-assisted process and guardrails I bring to your build.
Plansuru
Cost control & guardrailed AI
A group trip planner that ends group-chat chaos. Friends join by link with no account, share ideas, and vote on plans anonymously.
The Execution
- Cost guardrails: Map features run through one protected route with caching and limits, so usage can't run up the bill.
- Safe AI assistant: The built-in trip coach only sees what it should. Private notes stay private.
- Spec-first process: Written specs and automated tests guide every change, so AI-written code can't quietly break the app.
RoseS2
Automation & real-time sync
An app for couples to align on shared goals and household tasks, with everything synced in real time.
The Execution
- Hands-off automation: Daily nudges and notifications run on the server on a schedule, not on the phone.
- Living home screen: A hand-drawn garden that grows with the couple's progress, built entirely in code so it loads fast.
- Product-led AI development: AI writes the code; a central backlog, specs, and guardrails keep every session on direction.
Freezya
Safe releases & real-time multiplayer
Flashlight freeze tag in a dark, spooky neighborhood, played in the browser. Catch rivals in your beam to freeze them, then race to rescue your frozen teammates.
The Execution
- Smooth on any computer: An automatic graphics setting checks each player's machine and lowers detail when the game slows down.
- Updates never interrupt a game: New versions wait for matches in progress to finish, so players are never kicked mid-round.
- Guardrails on AI-written code: Automatic tests, written plans and a release lock stop unreviewed AI changes from reaching players.
Built by an Operator, Not a Code Gatekeeper.
I spent a decade running complex B2B ticketing operations and API integrations, where downtime and broken data cost real money.
Today I pair that operational discipline with Next.js, Supabase, and AI coding agents to work as a full engineering sprint team for founders. You get the speed of AI, plus someone who knows what it takes to keep a system running.
Operations meets engineering